Ai+ Training

When AI systems gain voice, presence, or physical representation, they don't just change what the system can do.

They change what humans do. People instinctively apply social rules to agents that project presence, hesitating to interrupt an avatar mid-sentence, lowering their voice when speaking to a voice agent, or placing trust in a system simply because it has a face and a voice. To the human brain, embodiment is not a design choice. It is a cognitive switch, one that restructures how users perceive, trust, and behave toward AI in ways that are automatic, measurable, and often irrational.

This keynote explores the intersection of cognitive science and system architecture, covering why embodiment hacks human cognition and the engineering realities required to support it, including the real-world trade-offs among latency, multimodal integration, and model intelligence.

The session includes a live technical walkthrough of building real-time voice and avatar agents using open-source tools and APIs, along with an analysis of embodiment's failure modes: uncanny discomfort, overtrust, and ethical risk, and how to catch them before production.

Whether you are choosing between text, voice, or a full avatar, you will leave with a new mental model for embodiment and the technical clarity to decide when to use it, when to avoid it, and how to build it right